The present disclosure relates generally to training an industrial robot. To teach a robot to perform intended tasks, a user currently has to use either offline software, such as Robot Studio, or an online device, such as a teach pendant, or a combination of both, to create a program, test it, and tune it. Existing robot training systems suffer from a number of shortcomings and disadvantages. There remain unmet needs including decreasing robot program creation, debugging, and testing complexity. For instance, robot training currently requires familiarity with robot programming language and software tools. Users must understand mathematically abstracted concepts, such as coordinate frames, and their transformation relations in order to effectively program the robot. Once a robot program is in service, it is difficult to troubleshoot since simulations of the robot program cannot be visualized in a real scene. There is a significant need for the unique apparatuses, methods, systems and techniques disclosed herein.